Trajectory evidence: what a session actually did, as against what it reported. Captures each seat's own transcript by path at SubagentStop, so an inspection cites the primary record rather than a summary of it.
Plugin gray-area from marketplace special-circumstances in ctoforaday/special-circumstances. Commands read from the repo's manifest — yesterday.
audit-checkpointPut the checkpoint's validation-loop claims against what this session actually ran. Reports CITED / STALE / NO-EVIDENCE / UNCHECKABLE with provenance on both sides. `--json` for rows, or pass a transcript path to audit a different session.
audit-pr-bodyAdjudicate a pull request body's claims against what the session actually ran. Reports CITED / NO-EVIDENCE / UNCHECKABLE with provenance on both sides, and names the parts of each claim the record cannot check. Exit is 0 even with findings.
audit-repetitionReport acts this session did more than once — the same file written again, the same command re-run — and acts repeated 3+ times BACK-TO-BACK, the [[anti-spinning]] limit adjudicated against the trajectory rather than the hook counter. `--json` for rows; pass a transcript path to inspect a different
audit-seat-coverageDoes the manifest name every seat transcript that exists? Reconciles the recorded seat rows against the session's own seat-transcript directory. Exits 1 when it could NOT measure, 0 when it did — whatever it found.
elicitation-testingWhen an agent or seat did something you cannot explain from its instructions — a rule it read and did not apply, a command run twice, a duty skipped — gather what it actually did with telepathy, then interview a tool-less fork of its own session about why, and adjudicate the answer against the recor
restart-recoveryAfter a host restart, crash or shutdown cut off running Claude sessions, use this to find them with `telepathy agents --lost` and bring each back in the mode it had — through a running Remote Control server first, then by reattach or local resume, one session at a time and only the ones the human ch
telepathyBefore asking a peer agent what happened, or assuming nobody else is working on a file, use `telepathy` to query the host-wide trajectory catalogue — it answers from what agents DID rather than from what they remember.
Core adversarial + cowork behaviour: Design-by-Contract rules-as-skills, pair programming, spec-driven development + a plan-audit gate, project memory, context checkpointing for compaction survival, proficiency skills, quality hooks, and environment preflight (/prosthetic-conscience:doctor).
27 skills · 2 agents
The research debate engine: an additive blue side, a subtractive red side that owns the pass/fail gate, an engine that runs the mechanics and a bench that rules the final compromise — best-of-N, a Heilmeier Catechism, and the full record preserved.
3 skills · 12 agents
Autonomous learning: a daily self-improvement loop, a graduation pipeline, and a continuous-learning promotion ladder — always human-gated at promotion.
Design fluency for frontend development. 1 skill with 24 commands (/impeccable polish, /impeccable audit, /impeccable critique, etc.) and curated anti-pattern detection.
1 skill · 4 agents · hooks
Agentic browser automation with Playwright for testing, scraping, and UI interaction
10 skills · 1 command · 1 agent
5 skills + 35 modes + Material Passport pipeline. Includes v3.6.7 cross-model audit gate and v3.6.8 generator-evaluator contract.
5 skills · 16 commands · 4 agents · hooks
Deep vulnerability scanning of your own code, run entirely inside your Claude Code session at a chosen effort tier, with every finding challenged before it is reported and the verification tally computed in code. Turns surviving findings into targeted patches, each verified by a panel of agents, that you apply when you choose.
1 skill · 9 agents · hooks
Independent auditing for AI agents, open source. Claude discovers your setup, confirms the cost, and audits your agent against the one question other tools skip: is it doing the job it's supposed to do, given your business rules and org structure? Test any model (Anthropic, OpenAI, Gemini, Azure, Bedrock, …) graded by self, one independent judge, or a cross-vendor panel; each provider is billed to its own account via a key in your settings, never the command line. Results render as an interactive artifact (with a static-report fallback).
1 skill · hooks
Cedar policy enforcement + Ed25519 signed receipts for every Claude Code tool call. First cryptographic governance plugin — decisions are policy-gated before they run and every decision produces a tamper-evident receipt verifiable offline.
1 skill · 2 commands · 2 agents · hooks
This page updates itself from your repo. Add the badge to your README, or get it Featured to pin it above the organic results with a followed link to your site.